EL34 - How Much Current Is Normal Under Hard Drive?

[Grantorino]

Hello again EL34 World.

Working on a Marshall 1987X Plexi 50W head.

Plate voltage around 460V. Bias set to approximately 35-36mA per tube at idle, as judged with a bias probe into a DMM. Probe sits between tube and tube socket.

Owner runs the amp cranked into a Fryette Power Station reactive attenuator.

When injecting a 400Hz 0.1V signal, V5 peaks at 133mA and V4 peaks at 160mA.

During guitar play-testing, V5 peaks around 160mA and V4 peaks around 180-200mA+ during sustained chord playing.

Both tubes recover quickly to idle current when playing stops. No red plating observed during short play-test.

How should these readings be interpreted? The asymmetry between the two sockets is consistent and does not change if the power tubes are switched between the sockets.

Are consistent peaks around 180-200mA inherently a problem?

Is the asymmetry in current between the V4 and V5 socket within normal parameters?

Work done:

The amplifier has new Belton power tube sockets, the PI coupling caps were replaced, as were the 5W screen resistors. All board-mounted resistors are within tolerance, the output transformer primary halves are balanced within 0.9 ohms, the power and pre-amp tubes were swapped out with no major change to asymmetry or high current peaks on the V4 socket.

At idle, the power tube bias point is stable. Is this behaviour under signal typical for an EL34-driven amplifier?

I would be interested to hear any thoughts.

Thank you.

[shooter]

it's "normal" for an amp to hit 2X power dissipation when driven deep into AB(2)
are you numbers "normal", not sure.  the way I test, I set my meter to Record Min/Max/Avg when driving hard, then do some quicky math to see how far "outta spec" it gets at Max n Avg, then re-adjust idle bias if it's running to hot outta spec using the AVG value math.


what Resistance is on your screens?

[pdf64]

It's all normal.
AB amps pass more current with signal than at idle, overdriven fixed bias it can be loads more.

The asymmetry is normal with Marshall.

I don't think any regular valve guitar amps operate in AB2?

[shooter]

AB2


I was in farmer mode when replying 


wasn't sure if this combo; 

Owner runs the amp cranked into a Fryette Power Station reactive attenuator.

would send one send it into unknown territory

[pdf64]

AB2


I was in farmer mode when replying 


wasn't sure if this combo; 

Owner runs the amp cranked into a Fryette Power Station reactive attenuator.

would send one send it into unknown territory

The uppermost anode characteristics plot usable in AB1 is Vg1=0. Above that, the control grid becomes forward biased and draws current from the signal source, clipping off the upper signal peaks unless the preceding driver stage can provide a high current signal.
Whereas in AB2, the driver stage is itself a power amp, that can provide lcurrent to pull the grid positive without the signal voltage becoming clipped.
Hence a key fundemental difference between AB1&2 is the driver stage, immediately preceding the output valves.
So however the anode is loaded, I can't see how it could affect whether AB2 operation is achievable?

[Merlin]

How should these readings be interpreted? The asymmetry between the two sockets is consistent and does not change if the power tubes are switched between the sockets.

The LTP is not perfectly balanced, especially during overdrive, so one of the power tubes gets driven a bit harder than the other. It's normal.

[HotBluePlates]

... Marshall 1987X Plexi 50W head.
...
When injecting a 400Hz 0.1V signal, V5 peaks at 133mA and V4 peaks at 160mA.

During guitar play-testing, V5 peaks around 160mA and V4 peaks around 180-200mA+ during sustained chord playing.

...

How should these readings be interpreted? ...

Are consistent peaks around 180-200mA inherently a problem?
...

A 50w Marshall's output transformer likely has a primary impedance around 3.4kΩ.

Power = Volts x Current, and
Volts = Current x Impedance, so
Power = (Current x Impedance) x Current = Current² x Impedance, so

RMS Current = √(Power / Impedance) = √(50w / 3.4kΩ) = 121.3mA RMS

The Peak Value of a (RMS) Sine Wave is Current RMS x √2, so
121.3mA RMS x 1.414 = ~172mA peak


Your 50w 1987X should manage 50 watts clean (and perhaps even a bit more), and will approach something like twice this much when distorted (2x sine-wave output power is the limit, which happens when the signal is distorted to symmetrical square-waves).

Your observations seem to be perfectly in-line with "clean signal test" when applying the 100mV 400Hz test signal, and "somewhat distorted signal" with your play-test.

[tubeswell]

Owner runs the amp cranked into a Fryette Power Station reactive attenuator.

These amps were originally designed not to run full bore permanently into power attenuators. That is extreme running. As others have said, it will cause the push-pull duty cycle between the LTP and the output tubes to become heavily lopsided.

[Grantorino]

it's "normal" for an amp to hit 2X power dissipation when driven deep into AB(2)
are you numbers "normal", not sure.  the way I test, I set my meter to Record Min/Max/Avg when driving hard, then do some quicky math to see how far "outta spec" it gets at Max n Avg, then re-adjust idle bias if it's running to hot outta spec using the AVG value math.


what Resistance is on your screens?

Thank you Shooter for the suggestion on using Min/Max/Avg recording, great tip. The screens are 1K 5W Wirewound, recently replaced. Screen voltage measurements around 447-452V at idle on each socket.

[Grantorino]

... Marshall 1987X Plexi 50W head.
...
When injecting a 400Hz 0.1V signal, V5 peaks at 133mA and V4 peaks at 160mA.

During guitar play-testing, V5 peaks around 160mA and V4 peaks around 180-200mA+ during sustained chord playing.

...

How should these readings be interpreted? ...

Are consistent peaks around 180-200mA inherently a problem?
...

A 50w Marshall's output transformer likely has a primary impedance around 3.4kΩ.

Power = Volts x Current, and
Volts = Current x Impedance, so
Power = (Current x Impedance) x Current = Current² x Impedance, so

RMS Current = √(Power / Impedance) = √(50w / 3.4kΩ) = 121.3mA RMS

The Peak Value of a (RMS) Sine Wave is Current RMS x √2, so
121.3mA RMS x 1.414 = ~172mA peak


Your 50w 1987X should manage 50 watts clean (and perhaps even a bit more), and will approach something like twice this much when distorted (2x sine-wave output power is the limit, which happens when the signal is distorted to symmetrical square-waves).

Your observations seem to be perfectly in-line with "clean signal test" when applying the 100mV 400Hz test signal, and "somewhat distorted signal" with your play-test.

Thank you HotBluePlates for doing the math on this and providing context for the readings I have taken. Knowing that a 172mA peak is expected is reassuring.

[Grantorino]

Observing the amplifier in low light, it has become clear that there is some red plating on the V4 tube (the same tube where I was seeing the highest peak current). The interesting observation is that when the Loudness 1 control is set fully open, rather than say 7-8, the red plating largely recedes. Regarding the intensity of the red plating - it is not severe and could only be seen in a dark room.

As I understand, at less than full volume there is more high frequency in the signal path. C5 is the 4n7 ‘bright cap’.

From the schematic (which I have attached), C15 appears to be a stability cap across the PI. It has not yet been tested and is a suspect.

Could high frequencies at less than full volume on channel 1 trigger a high frequency oscillation in the output stage, driving V4 to conduct excessively? Again, at full volume the 'bright cap' (C5) is effectively bypassed and the red plating mostly ceases.

Have anyone of you experienced gentlemen encountered this before?

EDIT - I should mention also that the channels are jumped in normal use, as is often done on the Plexi. Channel 1 and Channel 2 inputs connected with a short jumper cable.

Thanks for all the great contributions here.

[AlNewman]

If you're worried, you have the tools to monitor the plate and screen current/voltage for each tube under load.  If you have a stiffer PT, then under load you could be working the tubes past their maximum dissipation.

On the larger Fender amps, built for maximum clean power with solid state rectification, they often bias less than 50% at idle.  Sometimes even less than 40%.  I think that's because their transformers are a brick shithouse, and to bias higher would push the tubes into (eventual) detonation under load because there wouldn't be enough drop in input voltage as current increases.

If the amp is running with an attenuator, and is always pushing it's limits, I'd think this is even more important.

[Grantorino]

If you have a stiffer PT, then under load you could be working the tubes past their maximum dissipation.

On the larger Fender amps, built for maximum clean power with solid state rectification, they often bias less than 50% at idle. 

If the amp is running with an attenuator, and is always pushing it's limits, I'd think this is even more important.

Thank you Al.

During my previous diagnostic testing, I had tested different tube sets at a lower idle bias of around 28-30mA.

Comparing this to a higher idle bias of around 35-36mA, the peak current was actually measured to be lower on V4 (albeit with a signal generator):

Around 30mA at idle:
V5 peak - 118mA
V4 peak - 180mA
Asymmetry - 62mA

Around 35-36mA at idle:
V5 peak - 133mA
V4 peak - 160mA
Asymmetry - 27mA

V4 peaks as high as 200mA during sustained chord playing and quickly recovers to around 35mA at idle.

V5 peaks at approximately 160mA under the same conditions.

As previously suggested by Shooter, I think it would be wise to consider Min/Max/Avg rather than peak current.

I appreciate your comment about the power transformer possibly being stiff - I can check this - but the Loudness 1 set to 10 alleviating the red plating is still a question in my mind.

I will compare the B+ supply at idle vs under heavy signal, probably replace the 47pf cap and scope V4 and V5 grids to see if there is HF oscillation.

Thanks again.

[AlNewman]

Well, current doesn't actually mean much without voltages.
What happens if you lift the NFB?

[Grantorino]

I thought I would take some measurements and provide some data before lifting the NFB as AlNewman suggests.

First, a photo of the red plating in low-light conditions. This is while constantly strumming the guitar - it is not severe, but is certainly visible on V4 (right-hand tube) - see photo 1.

These test results show the voltage readings on pins 3, 4 and 5 of V4 at idle and under guitar signal.

      At idle          Under signal
Pin 3    447.5 V DC   381 V DC
Pin 4    440.8 V DC   348 V DC
Pin 5    -38.5 V DC   - 50 V DC

These results show there is actually significant voltage sag under signal.

Next I used the oscilloscope to take readings on the control grids of V4 and V5. I can provide some number if requested, but the readout was very similar - nearly identical Vp-p, nearly identical amplitude, similar frequencies. No obvious HF oscillation.

Testing the PI anodes, V3 pin 1:

• Freq: 827.91Hz
• Vmax: 260V
• Vmin: 112V
• Vp-p: 152V
• Amp: 144V
• Avg: 206V

V3 pin 6:

• Freq: 1.02KHz
• Vmax: 248V
• Vmin: 132V
• Vp-p: 114V
• Amp: 108V
• Avg: 180V

So I did my best to take readings while strumming the guitar. The numbers I posted here are from a screenshot of a recording of the readout - it seems that the two PI anodes are showing substantially different output (Pin 1 at 152V vs Pin 6 at 114V), and the average DC voltages differed by 26V.

How should these readings be interpreted? The asymmetry between the two sockets is consistent and does not change if the power tubes are switched between the sockets.

The LTP is not perfectly balanced, especially during overdrive, so one of the power tubes gets driven a bit harder than the other. It's normal.

So here Merlin clearly states the long-tailed pair are not balanced anyway, especially during overdrive. So perhaps this difference is expected. I think this explains why the V4 power tube is being driven harder.

I think next I should lift the NFB and evaluate the difference as AlNewman suggests.

Thanks again.

[Merlin]

If you change the LTP bias resistor from 470R to 1k you should find the power tube currents become more equal during overdrive.

[pdf64]

It may be that a degree of waveform asymmetry / uneven mark space ratio when heavily overdriven, is part of that classic sound.
It may indicate there's a fair bit of 2nd harmonic in there.

Whatever, I suggest to avoid probing output valve anodes, especially overdriven into a reactive load. The back emf spikes may run into several kV, unless protection diodes are fitted. Your meter might get wrecked (bitter experience  ).

[shooter]

^^^
what he said


I grab all my readings off the secondary "load", the math n scope images still work from there.....IF you do probe the anodes, use a 100X probe, something rated ~~ 5kv

[AlNewman]

Do you guys not measure dc plate voltages with a multimeter?

[shooter]

The back emf spikes may run into several kV, unless protection diodes are fitted. Your meter might get wrecked (bitter experience  ).

.....IF you do probe the anodes, use a 100X probe, something rated ~~ 5kv